nV News Forums

 
 

nV News Forums (http://www.nvnews.net/vbulletin/index.php)
-   Benchmarking And Overclocking (http://www.nvnews.net/vbulletin/forumdisplay.php?f=24)
-   -   ? about multiplier (http://www.nvnews.net/vbulletin/showthread.php?t=107028)

yxx 01-28-08 08:39 PM

? about multiplier
 
i'm pretty new at oc'ing. i have spent the last few days of looking at oc guides and i also read the article on andantech on the qx9650. i've encountered a lot of the "change multiplier or fsb" question. i want to start with a mild oc cause i want to keep my cpu for a long time.......so all i'm looking for is a 3.3 to 3.4 range for now and go higher later. i also bought a thermalright 120 ultra for my cpu.

if i change the mutiplier to 10 i get 3.3 (default 9)
if i change the fsb up enough i get also a 3.3range. but loose ram speed. (like 373mhz instead of 400)

so i guess my question is...is it damaging if i just change the multiplier to 10 and forget the rest..leave the fsb at 333 and keep my ram speed and ratio. i have ddr 2 800 and 4.4.4.12@2.1v. and qx9650 at stock for now.

i 3dmarked both and it's basically the same score.

if i go higher to like 3.6 and up can i just use mulitplier or is that dangerous and should i keep multiplier lower and use more fsb. thnx rest of specs in sig.

Uberpwnage 01-28-08 10:24 PM

Re: ? about multiplier
 
Generally a higher FSB will put more stress on the motherboard, so I would say just raise the multiplier, especially for mild overclocks. Of the QX9650 oc'es I've seen, alot of people go as high as a 12x multi, haven't seen much over that though. You should be able to get 3.9 or so on stock voltage or close to it, so theres not much harm in raising the multi. Alternatively, raising the FSB up to 400 shouldn't be difficult at all, however, once you reach the 430-440 range it may begin to get a little tough, depending on your setup. It's all personal preference really, neither option for such a mild oc is going to hurt your processor much, but I would assume raising the multi would be the safest of the two. Personal preference really.

Bman212121 01-28-08 11:41 PM

Re: ? about multiplier
 
When you raise the multipler you raise only the CPU speeds. If you raise the FSB you raise the CPU, the FSB, the memory, the PCIe / PCI slots etc. A lot of different components get their speed from the FSB. That being said, a lot of the higher end boards also allow you to change the divider to other components and which ones are linked to the FSB.

It can be a little confusing at first trying to figure out how everything relates to each other, but the reason there are so many different possibilities is that it makes it a lot easier to get the speeds you want. Having a CPU with an unlocked multiplier I would say go for the multiplier as well.

The one thing you will notice with the multipler is that it is done in large steps, so you may get to a point where you can no longer raise the multiplier because the gap is too large. This is where rasing the FSB a little can help get you some extra speed without losing stability. A lot of it is trial and error but what a lot of people do is figure out what the max speed they can run the CPU at by rasing the multiplier and writing that down. Then they take and lower the multipler below stock speeds (In your case 8x would be good) and then they increase the FSB until it becomes unstable. this way they can get an idea of how far they can push the FSB without the CPU causing problems. After you have those numbers then you can play around with the multiplier and the FSB and find a happy medium which gives the best speed.

mezkal 01-29-08 12:06 AM

Re: ? about multiplier
 
Personally I think you gain a lot more from increasing FSB speed than you do from just increasing the CPU clock speed. If your motherboard can fix PCI Buss to 100mhz whilst allowing you to increase the FSB you'll get a snappier machine overall - as long your RAM can handle it. That's a very general approach.

yxx 01-29-08 03:08 AM

Re: ? about multiplier
 
hey thanks for the responses and advice.........i think i understand now a bit more. i think if i am going to go higher than a 3.3ghz i will raise the fsb and go that way. like it was said above that a mild oc raising the multiplier one is pretty safe but raising the multiplier to 11 or 12 it takes big jumps so i think i will raise fsb and go that route higher when over 3.3ghz.

i never thought of seeing how high i can go stable with just multiplier and then also lowering multiplier and raising fsb and see how high i can go stable and then try to find a sweet spot.......i will do that for sure....i guess I'm a bit nervous raising the fsb to like 400 and adding more stress but i think i will find a happy medium. i play fsx so ill see how 3.3 does for now. i will go higher yet but i think i will look into getting another mobo this fall with ddr3.

mezkal what do you mean by a snappier machine in regards to raising the fsb?. like more frame rates in games or faster load up times or just general system feeling quicker. i guess i read around a bit that raising the fsb doesn't increase performance as much as people thought but i guess for higher clock speeds i will have to for better stability. sry for long post.

DRen72 01-29-08 08:39 AM

Re: ? about multiplier
 
I run 400x9 on my QX9650. I've found that the best compromise for speed and heat generation.
I can also run 400x10 completely stable but the heat gets a bit higher than I'd like when I run Prime95. Still stable though.

yxx 01-29-08 11:19 PM

Re: ? about multiplier
 
Quote:

Originally Posted by DRen72
I run 400x9 on my QX9650. I've found that the best compromise for speed and heat generation.
I can also run 400x10 completely stable but the heat gets a bit higher than I'd like when I run Prime95. Still stable though.

when you are running at 400x9 what are your temps when running prime 95. i am only running with the multiplier right now at 10 for 3.3 and i see you have the same heat sink fan as i.

it's pretty warm in my apartment but at 3.3 so far on full load core temp never shows any core higher than 45/47c. i will run like that for a bit and maybe try upping fsb and going for 3.6 this weekend and try that for a bit. also maybe its my case cooling but my mobo gets pretty warm at 3.3.......when play crysis a bit yesterday it got up to 38c at one point but my pc is in my living room and i have the fireplace on........so its warmer than usual so maybe that might be it.

DRen72 01-30-08 11:20 AM

Re: ? about multiplier
 
Quote:

Originally Posted by yxx
when you are running at 400x9 what are your temps when running prime 95. i am only running with the multiplier right now at 10 for 3.3 and i see you have the same heat sink fan as i.

it's pretty warm in my apartment but at 3.3 so far on full load core temp never shows any core higher than 45/47c. i will run like that for a bit and maybe try upping fsb and going for 3.6 this weekend and try that for a bit. also maybe its my case cooling but my mobo gets pretty warm at 3.3.......when play crysis a bit yesterday it got up to 38c at one point but my pc is in my living room and i have the fireplace on........so its warmer than usual so maybe that might be it.

When gaming at 400x9 the highest I've seen the CPU go was about 52c and that wasn't with Crysis. Crysis only managed to get up to around 48c.

When I stress tested the CPU, again at 400x9, using Prime95 it got up to 63c max. It went to 66c at 400x10. I didn't try 333x10, but I did try stock (333x9) with Prime95 and I recall the max temp being around 45c during testing.

However, I've not run across any program other than the stress tests that get the CPU over 52c.

Room temperature was also around 25-26c at the time which was rather warm.

yxx 01-31-08 07:10 PM

Re: ? about multiplier
 
Quote:

Originally Posted by DRen72
When gaming at 400x9 the highest I've seen the CPU go was about 52c and that wasn't with Crysis. Crysis only managed to get up to around 48c.

When I stress tested the CPU, again at 400x9, using Prime95 it got up to 63c max. It went to 66c at 400x10. I didn't try 333x10, but I did try stock (333x9) with Prime95 and I recall the max temp being around 45c during testing.

However, I've not run across any program other than the stress tests that get the CPU over 52c.

Room temperature was also around 25-26c at the time which was rather warm.

thanks for response..it's good to know.... I'm about on par with those temps too. when i ran stock cooler when i first got it it would hit about 60c on full load and i didn't want to go higher then that with oc ing cause i read that Intel max is like 64.5 for these cpu's....i would be more comfortable with a 400x9 with lower temps. thanks i will try both ways for a bit and see what that's like.

OH sry one more question are your temps different on each core on full load. i realize not all cores are always even loaded but i did some movie trans coding and all for cores where maxed out and there was a 6 degree difference between all four cores. core1 was like 47, core 2 41c core 3 43 and core 4 43c.

i also noticed this with the stock cooler too.

DRen72 02-01-08 09:59 AM

Re: ? about multiplier
 
Quote:

Originally Posted by yxx
...one more question are your temps different on each core on full load. i realize not all cores are always even loaded but i did some movie trans coding and all for cores where maxed out and there was a 6 degree difference between all four cores. core1 was like 47, core 2 41c core 3 43 and core 4 43c. I also noticed this with the stock cooler too.

Definitely. Typically my core1 and core2 are usually within a degree of each other, but core 3 & 4 can be about 4 degrees cooler. I say your results are normal for an unmodified Extreme 120 cooler.

yxx 02-02-08 01:43 AM

Re: ? about multiplier
 
Quote:

Originally Posted by DRen72
Definitely. Typically my core1 and core2 are usually within a degree of each other, but core 3 & 4 can be about 4 degrees cooler. I say your results are normal for an unmodified Extreme 120 cooler.

k thanks a lot for all your input.


All times are GMT -5. The time now is 07:33 PM.

Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.